Imported from Wang200935/openclaw-skills (
_merge_backups/auto_20260317_070002/misc/ai/ai-workflows-lcel/SKILL.md). Install upstream withnpx skills add Wang200935/openclaw-skills --skill ai-workflows-lcel. Copyright stays with the author.
LCEL-style Workflows
Use this skill when composing reusable AI pipelines.
Core model
Think in stages:
- input
- prompt shaping / transforms
- model or external API call
- post-processing
- output
Good defaults
- Keep each stage inspectable.
- Prefer explicit data contracts between stages.
- Make external API boundaries easy to stub or swap.
- Separate prompt construction from transport code.
- Log intermediate values for diagnosis, but avoid leaking secrets.
Use cases
- input -> prompt -> custom API -> parser -> UI
- file -> chunk -> retrieve -> synthesize
- status snapshot -> diagnose -> action plan -> terminal action
Read next when needed
references/patterns.mdfor chain patterns
MERGED_FROM_BACKUP: C:\Users\wang.openclaw\skills\ai\ai-workflows-lcel\SKILL.md
name: ai_workflows_lcel description: Design or reason about LangChain Expression Language style pipelines where input flows through transforms, models, tools, custom APIs, and outputs in a standard chain. Use when building chained AI workflows, wrappers around custom APIs, or structured prompt-to-output pipelines.
LCEL-style Workflows
Use this skill when composing reusable AI pipelines.
Core model
Think in stages:
- input
- prompt shaping / transforms
- model or external API call
- post-processing
- output
Good defaults
- Keep each stage inspectable.
- Prefer explicit data contracts between stages.
- Make external API boundaries easy to stub or swap.
- Separate prompt construction from transport code.
- Log intermediate values for diagnosis, but avoid leaking secrets.
Use cases
- input -> prompt -> custom API -> parser -> UI
- file -> chunk -> retrieve -> synthesize
- status snapshot -> diagnose -> action plan -> terminal action
Read next when needed
references/patterns.mdfor chain patterns